Using Machine Learning to Identify Alzheimer's Disease in the Initial Stages

This repository contains a TensorFlow-based image classification model designed for identifying different stages of Alzheimer's disease:

Model Overview

The model is built using TensorFlow and Keras, with a Sequential Model containing convolutional layers, max pooling, dropout, and dense layers. The final layer uses a softmax activation function for multi-class classification.

BinaryCrossentropy was chosen over CategoricalCrossentropy to improve the accuracy and efficiency of the model. Additionally, BinaryCrossEntropy takes classes independently and trains them, allowing it to go beyond classifying a Detected/Non-detected scenario.

Key Features:

• Data augmentation for training diversity
• Sequential model with Conv2D, MaxPooling2D, and Dense layers
• Softmax activation for multi-class classification
• SGD optimizer with learning-rate scheduler

Training Graphs

10 Epochs: ~80% Accuracy

• Training and validation accuracy and loss over 10 epochs.

These graphs represent the training and validation accuracy and loss over 10 epochs.

2 Epochs: ~78% Accuracy

• Training and validation accuracy and loss over 2 epochs.

Predicting Random Images from the Testing Data

Prediction Results

This section showcases the model's predictions on 8 random images from the testing data

Installation and Setup Instructions

Follow these steps to set up and run the detection model:

1. Clone the Repository

   git clone https://github.com/AdityaRao127/Alzheimers-ML-Detection.git
   cd Alzheimers-ML-Detection
   

2. Download the Datasets

   kaggle datasets download -d yasserhessein/dataset-alzheimer
   unzip dataset-alzheimer.zip -d path/to/dataset
   # replace the path with where you want to store your dataset. Example: "C:\Users\bob\Downloads\Alzheimers_Proj"
   

OR

Download manually from: https://www.kaggle.com/datasets/yasserhessein/dataset-alzheimer

3. Import the required libraries

   pip install tensorflow 
   pip install matplotlib 
   pip install numpy
   
   import tensorflow as tf
   from tensorflow import keras
   from tensorflow.keras.preprocessing.image import ImageDataGenerator
   import matplotlib.pyplot as plt
   from tensorflow.keras.models import Sequential
   from tensorflow.keras import layers
   from tensorflow.keras.optimizers import SGD
   from tensorflow.keras import models
   import numpy as np
   import matplotlib.pyplot as plt
   

4. Change paths

   # Change the testing and training directories appropriately, then run the cells in order.
   train_dir = "path/to/training/data"
   test_dir = "path/to/testing/data"
   
   

Credits

This project is developed as part of the research program at the Aspiring Scholars and Directed Research Program (May 2022 - Dec 2022)

License

This project is licensed under the MIT License - see the MIT LICENSE file for details.